移動通信大數據分析——數據挖掘與機器學習實戰

[中]歐陽曄(Ye Ouyang)[中]胡曼恬(Mantian Hu)[法]亞歷克西斯·休特(Alexis Huet)[中] 李中源(Zhongyuan Li)著,徐俊傑

立即出貨

  • 移動通信大數據分析——數據挖掘與機器學習實戰-preview-1
  • 移動通信大數據分析——數據挖掘與機器學習實戰-preview-2
  • 移動通信大數據分析——數據挖掘與機器學習實戰-preview-3
移動通信大數據分析——數據挖掘與機器學習實戰-preview-1

相關主題

商品描述

本書以4G/5G無線技術、機器學習和數據挖掘的新研究和新應用為基礎,對分析方法和案例進行研究;從工程和社會科學的角度,提高讀者對行業的洞察力,提升運營商的運營效益。本書利用機器學習和數據挖掘技術,研究移動網絡中傳統方法無法解決的問題,包括將數據科學與移動網絡技術進行完美結合的方法、解決方案和算法。 本書可以作為研究生、本科生、科研人員、移動網絡工程師、業務分析師、算法分析師、軟件開發工程師等的參考書,具有很強的實踐指導意義,是不可多得的專業著作。

目錄大綱

目 

目 

第1章概述

1.1 電信業大數據分析 ···························1

1.2 電信大數據分析的驅動力 ················2

1.3 大數據分析對電信產業價值鏈的

益處 ··················································3

1.4 電信大數據的實現範圍····················4

1.4.1 網絡分析 ···················································5

1.4.2 用戶與市場分析 ·······································8

1.4.3 創新的商業模式 ·······································91.5 本書概要 ··········································9

參考文獻 ·················································10

第2章電信分析方法論

2.1 回歸方法 ········································12

2.1.1 線性回歸 ··················································13

2.1.2 非線性回歸 ··············································15

2.1.3 特徵選擇 ··················································16

2.2 分類方法 ········································18

2.2.1 邏輯回歸 ··················································18

2.2.2 其他分類方法 ··········································19

2.3 聚類方法 ········································20

2.3.1 K均值聚類 ··············································21

2.3.2 高斯混合模型 ··········································23

2.3.3 其他聚類方法 ··········································24

2.3.4 聚類方法在電信數據中的應用 ·················25

2.4 預測方法 ········································25

2.4.1 時間序列分解 ··········································26

2.4.2 指數平滑模型 ··········································27

2.4.3 ARIMA模型 ············································28

2.5 神經網絡和深度學習 ·····················29

2.5.1 神經網絡 ··················································29

2.5.2 深度學習 ··················································31

2.6 強化學習 ········································32

2.6.1 模型和策略 ··············································33

2.6.2 強化學習算法 ··········································33

參考文獻 ·················································34

XII

XII

第3章 LTE網絡性能趨勢分析

3.1 網絡性能預測策略 ·························39

3.1.1 直接預測策略 ··········································39

3.1.2 分析模型 ··················································39

3.2 網絡資源與性能指標之間的關系 ···40

3.2.1 LTE網絡KPI與資源之間的關系 ···········40

3.2.2 回歸模型 ··················································41

3.3 網絡資源預測 ·································43

3.3.1 LTE網絡流量與資源預測模型 ···············43

3.3.2 預測網絡資源 ··········································43

3.4 評估RRC連接建立的應用 ············46

3.4.1 數據準備與特徵選取 ······························46

3.4.2 LTE KPI與網絡資源之間的關系推導 ····47

3.4.3 預測RRC連接建立成功率 ·····················49

參考文獻 ·················································50

第4章熱門設備就緒和返修率分析

4.1 設備返修率與設備就緒的預測

策略 ················································53

4.2 設備返修率和就緒預測模型 ··········54

4.2.1 預測模型的移動通信服務 ························54

4.2.2 參數獲取與存儲 ······································55

4.2.3 分析引擎 ··················································56

4.3 實現和結果 ·····································58

4.3.1 設備返修率預測 ······································58

4.3.2 設備就緒預測 ··········································62

第5章 VoLTE語音質量評估

5.1 應用POLQA評估語音質量··········68

5.1.1 POLQA標準···········································68

5.1.2 語音質量評價中的可擴展性和

可診斷性 ··················································69

5.2 CrowdMi方法論 ····························69

5.2.1 基於RF特徵的分類 ·······························70

5.2.2 網絡指標選擇與聚類 ······························70

5.2.3 網絡指標與POLQA評分之間的關系····70

5.2.4 模型測試 ··················································70

5.3 CrowdMi中的技術細節 ·················71

5.3.1 記錄分類 ··················································71

5.3.2 網絡指標的選擇 ······································71

5.3.3 聚類 ·························································72

5.3.4 回歸 ·························································73

5.4 CrowdMi原型設計與試驗 ·············74

5.4.1 客戶端和服務器架構 ······························74

5.4.2 測試和結果 ··············································76

參考文獻 ·················································78

 目 錄XIII

 目 錄XIII

第6章移動APP無線資源使用分析

6.1 起因和系統概述 ·····························80

6.1.1 背景和挑戰 ··············································80

6.1.2 移動資源管理 ··········································81

6.1.3 系統概述 ··················································82

6.2 AppWiR眾包工具 ··························83

6.3 AppWiR挖掘算法 ··························84

6.3.1 網絡指標的選擇 ······································84

6.3.2 LOESS方法 ············································87

6.3.3 基於時間序列的網絡資源使用預測 ·······87

6.4 實現和試驗 ·····································88

6.4.1 數據收集與研究 ······································88

6.4.2 結果和準確度 ··········································89

參考文獻 ·················································91

第7章電信數據的異常檢測

7.1 模型 ················································93

7.1.1 高斯模型 ··················································94

7.1.2 時間依賴的高斯模型 ······························94

7.1.3 高斯混合模型(GMM)·························95

7.1.4 時間依賴的高斯混合模型 ·······················95

7.1.5 高斯概率潛在語義模型(GPLSA)·······95

7.2 模型對比 ········································97

7.2.1 樣本定義 ··················································97

7.2.2 異常識別 ··················································98

7.2.3 時間依賴GMM與GPLSA的對比 ·········997.3 模擬與討論 ···································100

參考文獻 ···············································103

第8章基於大數據分析的LTE網絡自優化

8.1 SON(自組織網絡)···················105

8.2 APP-SON ······································107

8.3 APP-SON架構 ·····························108

8.4 APP-SON算法 ·····························110

8.4.1 匈牙利算法輔助聚類(HAAC)··········111

8.4.2 單位回歸輔助聚類數的確定 ·················114

8.4.3 基於DNN的回歸·································114

8.4.4 每個小區在時序空間的標簽組合 ·········116

8.4.5 基於相似性的參數調整 ·························1168.5 模擬與討論 ···································117

參考文獻 ···············································122

第9章電信數據和市場營銷

9.2.1 數據採集和數據類型 ····························130

9.1 電信營銷專題 ·······························127

9.2.2 網絡的提取和管理 ································131

9.2 社交網絡的總體構建 ···················130

9.3 網絡結構的度量 ···························133

參考文獻 ···············································135

9.4 網絡中的消費者行為建模 ············134

第10章傳染式客戶流失

10.1 問題引入 ·····································138

10.1.1 流失率問題 ··········································138

10.1.2 社交學習和網絡效應 ··························139

10.2 網絡數據的處理 ·························141

10.3 動態模型 ·····································143

10.3.1 模型介紹 ··············································143

10.3.2 模型的定義 ··········································144

10.3.3 自身經驗建模、社交學習和

社交網絡效應 ······································146

10.3.4 模型估計 ··············································148

10.4 結果 ············································149

參考文獻 ···············································151

第11章基於社交網絡的精準營銷

11.1 網絡效應的渠道 ·························158

11.2 社交網絡數據處理 ·····················159

11.3 建模策略問題 ·····························160

11.3.1 線性空間自回歸模式 ···························160

11.3.2 社交網絡交互模型 ······························162

11.3.3 內生同伴效應 ······································162

11.4 發現與應用 ·································164

11.4.1 結果的解釋 ··········································164

11.4.2 基於社交網絡的精準營銷 ···················165

參考文獻 ···············································168

第12章社交影響和動態社交網絡結構

12.1 動態模型 ·····································17712.1.1 連續時間馬爾可夫模型假設 ···············17712.1.2 模型估計與識別 ··································17912.1.3 網絡結構對社交影響的多元分析 ·······18012.2 研究發現總結 ·····························18112.2.1 隨機行動者動態網絡模型的

估計結果··············································182

12.2.2 元回歸分析結果 ··································184

12.2.3 策略模擬 ··············································18812.3 結論 ············································193

參考文獻 ···············································194